V. LaserWel: A Defect Detection and Monitoring System for Laser Welding
5.2 Software development
Table 5-4 Detailed specifications of filters for the sensors system
Optical parts
Transmission band (nm)
Cutoff wavelength
(nm)
Graph
ND filter 350-2700 350
Longpass filter
(for temperature sensor) 1100-2200 1100
Beam splitter 685 - 1600 650
Broadband mirror 200-2200 2600
Each stitch is analyzed by selected detection method. Defect detection result, and then, are listed in the right bottom table with the different background color of cells(good weld: green, bad weld: red).
The specific defect detection information on each stitch are shown when the user double clicked the specific cell in the table.
As the software is connected to the main database, the end-user should log on the database using
“ Login” menu. Basically, all the data is managed by creating a “Project workspace”. The project, first, should include the production(experiment) date, part specification.
Using the previous experimental data, one can conduct the off-line analysis. The previous data can be loaded from either database or external CSV files. The signal loaded is visualized on the graph window. The signal automatically classifies each stitch. Based on the condition, the off-line analysis method is selected among the reference curve(univariate) generation, defect features, or pattern model(multivariate) generation. Trained reference curves are overlapped and visualized on the right graph window(see Figure 5-6). All the training information and the results belong to the current project workspace. The project can have several off-line training results.
Figure 5-5 A main panel of the developed monitoring software
Figure 5-6 An off-line training panel of the developed monitoring software
Figure 5-7 A reference selection panel for on-line monitoring of the developed monitoring software
Before the on-line monitoring, an end-user can select one of the previous trained off-line analysis results. Weld defect method is selected based on which off-line analysis is used for training. If training conducted both univariate reference curve generation or thresholds, one could select either of them as a defect detection method. Initial welding parameters such as laser power and welding speed(feed rate) are used for further gap assessment detection criteria.
Figure 5-8 Another version of developed monitoring software for production environment
A compact version of the module is sometimes needed in the case of the production system, even though the developed software is well-functioned and user-centric. For the production environment, we developed another version of the software which is called LaserWel as shown in Figure 5-8.
The main function of this software is focusing on the monitoring the current welding process and defect detection only. Off-line analysis function, database and project management function are removed for compactness and easiness as an aspect of field workers.
The project and trained off-line analysis results from the main software above can be selected by the production software(LaserWel) handler, the total number of the weld, the detected defects, and its specific logs are provided.
This Page Intentionally Left Blank
As the use of laser welding application increases so does the needs for reliable quality monitoring methods. For this purpose, attention should be focused on defect detection on the quality estimation of laser welding. Many systems have used statistical approaches for fault detection. There were still limitations in applicability on signals which had a trend. The Dempster-Shafer theory based defect detection method can be advantageously used to overcome the limitations in laser welding process monitoring.
The weld pool temperature, plasma intensity, and back reflected laser signal were gathered regarding the different radiation wavelengths. Their characteristics of nominal trends are estimated by probability distribution function estimation to extract thresholds. We proposed a modified probability assignment function with respect to the thresholds. The weld defect detection of each information source(sensors) was processed respectively. We then aggregated the individual evidence of normal state were fused by using the combination rule of the Dempster-Shafer theory. The performance of the developed detection method is evaluated by statistical comparison of visual inspection result(real defect) and detection result. The result implies that the fusion of the reliable sensors increased the accuracy of defect detection. The defect detection method was eventually embedded in the monitoring software system.
In this thesis, we developed a process monitoring system for laser welding including (i) a hardware configuration of the data acquisition system, (ii) a sensor fusion-based defect detection algorithm for laser welding, and (iii) an easy-to-use GUI-based software that is an essential feature for industrial usage.
We defined the problem or challenging issues on the defect detection of the multi-sensory system.
First, the final decision or classification of the target state cannot be simply achieved by the multiple sensor based monitoring system. It is possible that the individual sensors indicated different decisions in the same state. We adopted the combination rule of the Dempster-Shafer theory as a sensor fuser of the individual defect decisions.
Second, even if all the sensor signals were within the tolerance ranges, the actual defect may have happened in a practical environment. This is an in-tolerance failure problem. We proposed a modified
probability assignment function to adopt the in-tolerance defects. By controlling the slope of the probability assignment function, we were able to treat and assign the uncertainty to the in-tolerance signal at some degree of defect evidence.
The last problem that we did not focus on this thesis was the specific pattern of defect signal that might exist. In this case, we need to model the time-series signal pattern or to choose the number of pieces of segments(called as a binning problem). We also need methods to codify the segmented pieces called a codification problem and to interpret the series of codified segments which is related to the process mining problem.
We further extended the defect detection and monitoring process to the process adjustment. It is a concept of a self-resilient control system which can control the quality automatically. The system appropriately adjusts the laser welding process parameters such as laser power, welding speed, welding direction, in order to recover the current laser welding process from the faulty state. Naturally, statistical analysis of welding experiments must be made a priori in a view to constructing a response surface that proposes appropriate adjust strategies including (i) identification of laser welding process parameters to be adjusted, and (ii) their optimal adjustment magnitudes with respect to the identified welding conditions, especially part-to-part gap. To do this, we will develop a part-to-part gap assessment method to determine the in-process welding condition in a real-time manner. The pilot experiments for the gap assessment model was listed in Appendix. Several regression and classification algorithms such as a support vector machine and neural network will be extended to weld signal trend clustering. By using the results of the part-to-part gap assessment and the associated adjustment strategies, it is possible to expect a better welding quality for the next batches at the same weld part batch of the assembly operation.
Finally, the defect detection module for laser welding, the constructed response surface, and the online part-to-part gap assessment module will be integrated as a closed-loop laser welding process controller. Validation and verification of the developed modules and methods will be carried out.
1. R. Oh, D. -Y. Kim, and D. Ceglarek, "The effects of laser welding direction on joint quality for non-uniform part-to-part gaps," Metals, vol. 6, no. 8, p. 184, 2016.
2. K. Benyounis and A. Olabi, "Optimization of different welding processes using statistical and numerical approaches–A reference guide," Advances in Engineering Software, vol. 39, no. 6, pp. 483-496, 2008.
3. K. Benyounis, A. Olabi, and M. Hashmi, "Effect of laser welding parameters on the heat input and weld-bead profile," Journal of Materials Processing Technology, vol. 164, pp. 978-985, 2005.
4. J. Shao and Y. Yan, "Review of techniques for on-line monitoring and inspection of laser welding," Journal of Physics: Conference series, vol. 15, no. 1, p. 101, 2005.
5. H. Gu and W. Duley, "A statistical approach to acoustic monitoring of laser welding," Journal of Physics D: Applied Physics, vol. 29, no. 3, p. 556, 1996.
6. H. Zeng, Z. Zhou, Y. Chen, H. Luo, and L. Hu, "Wavelet analysis of acoustic emission signals and quality control in laser welding," Journal of Laser Applications, vol. 13, no. 4, pp. 167-173, 2001.
7. L. Li, "A comparative study of ultrasound emission characteristics in laser processing," Applied Surface Science, vol. 186, no. 1, pp. 604-610, 2002.
8. Y.-W. Park, "Weld process monitoring technology in laser welding," Journal of Welding and Joining, vol. 30, no. 1, pp. 27-32, 2012.
9. R. Olsson, I. Eriksson, J. Powell, A. Langtry, and A. Kaplan, "Challenges to the interpretation of the electromagnetic feedback from laser welding," Optics and Lasers in Engineering, vol.
49, no. 2, pp. 188-194, 2011.
10. I. Eriksson, J. Powell, and A. Kaplan, "Signal overlap in the monitoring of laser welding,"
Measurement Science and Technology, vol. 21, no. 10, p. 105705, 2010.
11. Y. W. Park, H. Park, S. Rhee, and M. Kang, "Real time estimation of CO2 laser weld quality for automotive industry," Optics & Laser Technology, vol. 34, no. 2, pp. 135-142, 2002.
12. T. Sibillano, A. Ancona, V. Berardi, and P. Lugara, "Correlation analysis in laser welding plasma," Optics Communications, vol. 251, no. 1, pp. 139-148, 2005.
13. J. F. Tu, T. Inoue, and I. Miyamoto, "Quantitative characterization of keyhole absorption mechanisms in 20 kW-class CO2 laser welding processes," Journal of Physics D: Applied Physics, vol. 36, no. 2, p. 192, 2002.
14. Y. Zhang, L. Li, and G. Zhang, "Spectroscopic measurements of plasma inside the keyhole in deep penetration laser welding," Journal of Physics D: Applied Physics, vol. 38, no. 5, p. 703, 2005.
15. F. Bardin, A. Cobo, J. M. Lopez-Higuera, O. Collin, P. Aubry, T. Dubois, M. Högström, P. Nylen, P. Jonsson, and J. D. Jones, "Optical techniques for real-time penetration monitoring for laser welding," Applied Optics, vol. 44, no. 19, pp. 3869-3876, 2005.
16. D. You, X. Gao, and S. Katayama, "Review of laser welding monitoring," Science and Technology of Welding and Joining, vol. 19, no. 3, pp. 181-201, 2014.
17. H. Park and S. Rhee, "Analysis of mechanism of plasma and spatter in CO2 laser welding of galvanized steel," Optics and Laser Technology, vol. 31, no. 2, pp. 119-126, 1999.
18. D. Colombo and B. Previtali, "Fiber laser welding of titanium alloys and its monitoring through the optical combiner," proceedings of International Congress on Applications of Lasers &
Electro-Optics, pp. 620-629, 2009
19. M.-K. Jeon, W.-B. Kim, G.-C. Han, and S.-J. Na, "A study on heat flow and temperature monitoring in the laser brazing of a pin-to-plate joint", Journal of Materials Processing Technology, vol. 82, no. 1, pp. 53-60, 1998.
20. C. Kim, J. Kim, H. Lim, and J. Kim, "Investigation of laser remote welding using disc laser", Journal of Materials Processing Technology, vol. 201, no. 1, pp. 521-525, 2008.
21. ISO, BSEN. 13919-1: 1997, "Welding-Electron and laser beam welded joints-Guidance on
quality levels for imperfections-Part 1", 1996.
22. ISO, EN. 13919-2: 2001, "Welding-Electron and laser beam welded joints–Guidance on quality levels for imperfections–Part 2", 2001.
23. J. Beersiek, "A CMOS camera as a tool for process analysis not only for laser beam welding", International Congress on Applications of Lasers & Electro-Optics, 2001.
24. J. Beersiek, "On-line monitoring of keyhole instabilities during laser beam welding", International Congress on Applications of Lasers & Electro-Optics, vol. 99, 1999.
25. Y. Zhang, R. Kovacevic, and L. Li, "Characterization and real-time measurement of geometrical appearance of the weld pool", International Journal of Machine Tools and Manufacture, vol.
36, no. 7, pp. 799-816, 1996.
26. C. Brock, F. Tenner, F. Klämpfl, R. Hohenstein, and M. Schmidt, "Detection of weld defects by high speed imaging of the vapor plume", Physics Procedia, vol. 41, pp. 539-543, 2013.
27. H. Luo and X. Chen, "Laser visual sensing for seam tracking in robotic arc welding of titanium alloys", The International Journal of Advanced Manufacturing Technology, vol. 26, no. 9, pp.
1012-1017, 2005.
28. W. Huang and R. Kovacevic, "A laser-based vision system for weld quality inspection", Sensors, vol. 11, no. 1, pp. 506-521, 2011.
29. W. Huang and R. Kovacevic, "Development of a real-time laser-based machine vision system to monitor and control welding processes", The International Journal of Advanced Manufacturing Technology, vol. 63, no. 1-4, pp. 235-248, 2012.
30. V. Kaftandjian, O. Dupuis, D. Babot, and Y. M. Zhu, "Uncertainty modelling using Dempster–
Shafer theory for improving detection of weld defects", Pattern Recognition Letters, vol. 24, no. 1, pp. 547-564, 2003.
31. N. Nacereddine, M. Zelmat, S. Belaifa, and M. Tridi, "Weld defect detection in industrial radiography based digital image processing", Transactions on Engineering Computing and Technology, vol. 2, pp. 145-148, 2005.
32. C.-J. Lee, J.-D. Kim, and Y.-C. Kim, "Study on monitoring of plasma emission signal in lap welding of Zn coated steel sheet using CO2 laser", International Journal of Precision Engineering and Manufacturing, vol. 16, no. 3, pp. 495-500, 2015.
33. J.-D. Kim and C.-J. Lee, "Study on the Relationship Between Emission Signals and Weld Defect for In-Process Monitoring in CO2 Laser Welding of Zn-Coated Steel", Transactions of the Korean Society of Mechanical Engineers A, vol. 34, no. 10, pp. 1507-1512, 2010.
34. D. H. Kim, H. J. Shin, and Y. T. Yoo, "A study on the digital filter and wavelet transform of monitoring for laser welding", Journal of the Korean Society for Precision Engineering, vol.
30, no. 1, pp. 67-76, 2013.
35. S. Saludes Rodil, R. Arnanz Gómez, J. M. Bernárdez, F. Rodríguez, L. J. Miguel, and J. R.
Perán, "Laser welding defects detection in automotive industry based on radiation and spectroscopical measurements", The International Journal of Advanced Manufacturing Technology, vol. 49, no. 1, pp. 133-145, 2010.
36. D. Bebiano and S. C. Alfaro, "A weld defects detection system based on a spectrometer", Sensors, vol. 9, no. 4, pp. 2851-2861, 2009.
37. H. Luo, H. Zeng, L. Hu, X. Hu, and Z. Zhou, "Application of artificial neural network in laser welding defect diagnosis", Journal of Materials Processing Technology, vol. 170, no. 1, pp.
403-411, 2005.
38. I. Eriksson and A. Kaplan, "Evaluation of laser weld monitoring: a case study", International Congress on Applications of Lasers & Electro-Optics, vol. 102, pp. 1419-1425, 2009.
39. D. Colombo, B. M. Colosimo, and B. Previtali, "Comparison of methods for data analysis in the remote monitoring of remote laser welding", Optics and Lasers in Engineering, vol. 51, no.
1, pp. 34-46, 2013.
40. Y. Li, Y. F. Li, Q. L. Wang, D. Xu, and M. Tan, "Measurement and defect detection of the weld bead based on online vision inspection", IEEE Transactions on Instrumentation and Measurement, vol. 59, no. 7, pp. 1841-1849, 2010.
41. P. Zhang, L. Kong, W. Liu, J. Chen, and K. Zhou, "Real-time monitoring of laser welding based on multiple sensors", Control and Decision Conference, pp. 1746-1748, 2008.
42. A. Sun, E. Kannatey-Asibu Jr, and M. Gartner, "Monitoring of laser weld penetration using sensor fusion", Journal of Laser Applications, vol. 14, no. 2, pp. 114-121, 2002.
43. M. Kuhl and R. Neugebauer, "Quality and process control of Nd: YAG-laser welding using fuzzy pattern recognition techniques for multisensor systems", IEEE Conference on Emerging Technologies and Factory Automation, pp. 1329-1332, 2006.
44. O. Basir and X. Yuan, "Engine fault diagnosis based on multi-sensor information fusion using Dempster–Shafer evidence theory", Information Fusion, vol. 8, no. 4, pp. 379-386, 2007.
45. J. Ma, F. Kong, W. Liu, B. Carlson, and R. Kovacevic, "Study on the strength and failure modes of laser welded galvanized DP980 steel lap joints", Journal of Materials Processing Technology, vol. 214, no. 8, pp. 1696-1709, 2014.
46. W. Chen, P. Ackerson, and P. Molian, "CO2 laser welding of galvanized steel sheets using vent holes", Materials & Design, vol. 30, no. 2, pp. 245-251, 2009.
47. L. Mei, D. Yan, J. Yi, G. Chen, and X. Ge, "Comparative analysis on overlap welding properties of fiber laser and CO 2 laser for body-in-white sheets", Materials & Design, vol. 49, pp. 905- 912, 2013.
48. L. Mei, G. Chen, X. Jin, Y. Zhang, and Q. Wu, "Research on laser welding of high-strength galvanized automobile steel sheets", Optics and Lasers in Engineering, vol. 47, no. 11, pp.
1117-1124, 2009.
49. D. Westerbaan, D. Parkes, S. Nayak, D. Chen, E. Biro, F. Goodwin, and Y. Zhou, "Effects of concavity on tensile and fatigue properties in fibre laser welding of automotive steels", Science and Technology of Welding and Joining, vol. 19, no. 1, pp. 60-68, 2014.
50. A. K. Sinha, D. Y. Kim, and D. Ceglarek, "Correlation analysis of the variation of weld seam and tensile strength in laser welding of galvanized steel", Optics and Lasers in Engineering, vol. 51, no. 10, pp. 1143-1152, 2013.
51. R. Oh and D. -Y. Kim, "A fault detection method of laser welding based on PDF estimation and Dempster-Shafer theory", Proceedings of the Society of CAD/CAM Engineers Conference, 2014, pp. 1011-1016.
52. G. Shafer, "Dempster-Shafer theory", Encyclopedia of Artificial Intelligence, pp. 330-331, 1992.
53. K. Sentz and S. Ferson, "Combination of evidence in Dempster-Shafer theory", Sandia National Labs., Albuquerque, NM (US); Sandia National Labs., Livermore, CA (US)2002.
54. P. Smets, "Belief functions: the disjunctive rule of combination and the generalized Bayesian theorem", International Journal of Approximate Reasoning, vol. 9, no. 1, pp. 1-35, 1993.
55. Q. Chen and U. Aickelin, "Anomaly detection using the Dempster-Shafer method", SSRN:
http://dx.doi.org/10.2139/ssrn.2831339, 2006.
56. M. Goldstein and S. Uchida, "A comparative evaluation of unsupervised anomaly detection algorithms for multivariate data", PloS One, vol. 11, no. 4, p. e0152173, 2016.
57. C. Siaterlis and B. Maglaris, "Towards multisensor data fusion for DoS detection", Proceedings of the 2004 ACM symposium on Applied computing, 2004, pp. 439-446: ACM.
58. G. Shafer, "Dempster's rule of combination", International Journal of Approximate Reasoning, vol. 79, pp. 26-40, 2016.
59. G. Shafer, "A mathematical theory of evidence", Princeton University Press, 1976.
60. C. Thomas and N. Balakrishnan, "Advanced sensor fusion technique for enhanced intrusion detection", IEEE International Conference on Intelligence and Security Informatics, pp. 173- 178, 2008.
61. A. Patcha and J.-M. Park, "An overview of anomaly detection techniques: Existing solutions and latest technological trends", Computer Networks, vol. 51, no. 12, pp. 3448-3470, 2007.
Acknowledgments
I would like to express my sincere gratitude to my advisor Professor Duck-Young Kim for the continuous support of my study, for his patience, motivation, and immense knowledge. The door of his office was always open whenever I ran into a trouble and had a question about my research or writing.
He consistently allowed this paper to be my own work, but steered me in the right the direction. Besides my advisor, I would like to thank the rest of my thesis committee: Prof. Daeil Kwon and Prof. Hyungson Ki for their insightful comments and encouragement, but also for the hard question which incented me to widen my research from various perspectives.
I would like to thank my fellow labmates in for the stimulating discussions, for the sleepless nights we were working together before deadlines, and for all the fun we have had.
Special thanks to going to Sungwoo Hitech Co. Ltd. for supporting the experiments with the 6.6kw disk laser welding system. and also thanks to the financial supports by the international collaborative R&D program of Korea Institute for Advancement in Technology (Grant no. EUFP-M0000224) and the EU FP7 project of the European Commission (Grant no. FP7 Project 285051).
Finally, I must express my very profound gratitude to my family and to my girlfriend for providing me with unfailing support and continuous encouragement throughout for many years. This accomplishment would not have been possible without them.
Thank you.
Appendix
A. Experiments for the part-to-part gap size
Figure A-1 A trend of the plasma signal and temperature signal on the part-to-part gap size 15
20 25 30 35 40
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
Plasma signal (V)
Time (s)
0.05 mm 0.1 mm 0.15 mm 0.2 mm 0.25 mm 0.3 mm
12 14 16 18 20 22
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
Temperature signal (V)
Time (s)
0.05 mm 0.1 mm 0.15 mm 0.2 mm 0.25 mm 0.3 mm
Figure A-2 A simple regression model (light intensity vs. part-to-part gap)
Table A-1 Partial fraction factorial design including center points with three process parameters at two-levels
Code unit Experimental factors
Laser power (W) Welding speed (mm/min)
Part-to-part gap (mm)
-1 1600 700 0.05
1 2000 1100 0.25
Center point: Laser power = 1800W, Welding speed = 900mm/min, Part-to-part gap = 0.15mm
Table A-2 The experimental data for experiment of part-to-part gap assessment
Experimental factors Responses
Laser power (W)
Welding speed (mm/min)
Part-to-part gap (mm)
Plasma (V)
Temperature (V)
Reflection (V)
2000 700 0.05 6.348 6.479 2.266
2000 1100 0.25 3.925 13.258 2.278
1600 1100 0.05 3.520 3.989 1.810
2000 700 0.05 6.648 7.065 2.271
1800 900 0.15 3.809 11.256 2.048
1800 900 0.15 3.591 12.373 2.045
1600 1100 0.05 3.658 4.090 1.807
1600 700 0.25 4.082 10.658 1.810
2000 1100 0.25 4.269 13.572 2.273
Plasma = -31.943*Gap + 25.404 R² = 0.6825
Temperature = -19.727*Gap + 17.625 R² = 0.9809
10 13 16 19 22 25
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Voltage(V)
Part-to-part gap (mm) Plasma Temperature
regression model(Plasma) regression model(Temperature)
2000 700 0.05 6.337 7.531 2.270
1800 900 0.15 3.417 11.119 2.051
1800 900 0.15 3.643 11.597 2.055
1600 700 0.25 3.460 10.798 1.811
2000 1100 0.25 4.985 11.388 2.272
1800 900 0.15 4.552 11.122 2.052
1600 700 0.25 4.697 10.451 1.809
1600 1100 0.05 3.723 4.211 1.808
2000 700 0.05 6.348 6.479 2.266
2000 1100 0.25 3.925 13.258 2.278
1600 1100 0.05 3.520 3.989 1.810
2000 700 0.05 6.648 7.065 2.271
Table A-3 ANOVA table of response surface regression for the experiment of part-to-part gap assessment
Source Degree of
freedom
Sum of squares
Mean
square F-ratio P-value
Laser power 1 25.047 25.047 80.03 0
Welding speed 1 0.373 0.373 1.19 0.291
Part-to-part gap 1 151.479 151.479 483.98 0
Laser power * Laser power 1 31.107 31.107 99.39 0
Error 16 5.008 0.313
Total 20 222.419
R-square = 0.97775
Temperature(V) = -236.3 + 0.2646ⅹ Laser power(W) + 31.28ⅹPart-to-part gap(mm) - 0.000072ⅹLaser power(V)ⅹLaser power(V)
Figure A-3 The estimated part-to-part gap size by the response surface model
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Gap 0.05 0.25 0.05 0.05 0.15 0.15 0.05 0.25 0.25 0.05 0.15 0.15 0.25 0.25 0.15 0.25 0.05 Estimated_gap 0.032 0.267 0.046 0.051 0.142 0.179 0.05 0.251 0.277 0.067 0.138 0.153 0.255 0.206 0.138 0.244 0.054
0 0.05 0.1 0.15 0.2 0.25 0.3
Part to part gap (mm)